8th European Breast Cancer Conference (EBCC-8), March 21 - 24, Vienna, Austria

Mouse models of metastatic breast cancer

Professor Jos Jonkers - Netherlands Cancer Institute, Amsterdam

I gave the first talk of the session on mouse models for metastatic disease, in fact the entire session dealt with metastatic breast cancer. My talk focused on mouse models for metastatic breast cancer and I specifically talked about the newer mouse models that we’ve been developing, among other labs, where we take information from human cancers and, via genetic engineering, introduce this in mice in order to then predispose them to developing cancer as it occurs in patients but also to the extent that they will develop then metastatic disease.

The basis of our models is P3 mutated breast cancer, and the reason is quite simple: the majority of breast cancers and solid cancers, of course, in patients are P3 mutated. These mice develop, when you inactivate P3 in a tissue-specific fashion, mammary tumours that are non-invasive and non-metastatic. So this allowed us to then ask if additional mutations and other tumour suppressor genes would then confer a metastatic phenotype and indeed we found that when we induced mutations or co-mutation of E-cadherin that then the tumours became metastatic. And this links to the observation that E-cadherin loss in vitro, in cell lines, induces increased motility and an in vitro invasion phenotype. And it also links to the fact that the breast cancers that are hallmarked by E-cadherin mutation, lobular breast cancer in patients, are hallmarked by typical late term metastasis that may occur up to ten years after the local treatment has been performed.

Do you have specific models for specific types of metastases?

We don’t. I would argue that in terms of developing models for spontaneous tumour metastasis, it’s still relatively early days, so most of the tumour models are good at mimicking or modelling the early stages of tumour development; fewer models are good at mimicking the later stages. So we have this model that links to lobular breast cancer in women, the E-cadherin mutated model. We more recently developed, in fact this was more developed by another group in the Institute, the group of Anton Berns, a different model where metastatic breast cancer is induced by co-mutation of BP3 and DCC, that’s another tumour suppressor gene, so in that sense we’re making let’s say slow progress in building up a larger array of models for metastatic breast cancer, but it’s a slow process.

There are two issues. One is heterogeneous disease, and of course that means that there is also heterogeneity in terms of the way that metastases build up. In order to really be able to view and document metastatic disease development, you also have to get past the stage where animals succumb to their primary tumour, so in the case of this E-cadherin mutated breast cancer model I think we’re lucky in the sense that it has quite a high propensity to spontaneously metastasise. In other cases where the chance of building up metastatic disease is slower, of course mice will succumb to the primary tumour, and then you need to really implement a clinical protocol where you do local regional treatment, and then wait until the animals may succumb to metastatic disease.

What’s the plan and eventual aim for this research?

I think two issues: one a deeper understanding of metastatic breast cancer, and the point here is that metastasis is really a process that is governed by a cascade of steps, so cells need to detach from the primary tumour, they need to extravasate, they need to enter either the bloodstream or the lymph system, and then of course they need to also exit these systems again, at a distant site, where they need to build up a niche. This cascade is still, to a large extent, pretty obscure, at least the mechanisms that underlie all of these steps in the cascade, so a greater fundamental knowledge is one goal, I think, that we are trying to achieve.

And the other issue is really that we want to cure patients in the metastatic setting. It’s clear, especially in breast cancer, that survival greatly decreases when patients really cross that, as the last speaker in the session mentioned, that purple line where tumours really disseminate and have grown detectable metastases. So we want to be able to also offer therapies to these patients, and then you need models that really also adequately recapitulate these stages of the disease in order to really identify candidate drug targets, test the validity of these targets using single drug treatments or combinatorial strategies. I think this is really a slow process. I think that the examples where targeted therapeutics really have worked are the examples that we are all aware of, and in most of the cases the mouse models have not been able, or not really made a significant contribution there. But we also know that the majority of metastatic cancers can still not be treated, so we have now a few examples where targeted therapeutics have made a difference and have indeed shown stable durable responses. We have more examples where targeted therapeutics do nothing, or give very temporal responses, rapid induction of resistance, etc. So it’s clear that we need to learn a lot, that we need good models in order to go through these iterative cycles of learning, testing, learning, testing.

Are we at the start of a large increase in targeted therapeutics?

Yes, although I think that it’s going to be a slow gain, and that in many cases we need to tailor our level of combinatorial strategies, so I think that in many cases we see that the example of Imatinib and CML or GST is more an exception than a rule and that you see addiction, oncogene addiction, is a very temporal trait, a kind of 'temporal slavery’, as we once wrote in a review, and cells quickly re-wire to become resistant. And in other cases it turns out that the candidate target is irrelevant and that some of these lesions are relevant for tumour initiation but not for tumour maintenance.